高光谱遥感蚀变矿物填图算法对比研究及应用

发布时间：2018-03-27 09:00

本文选题：高光谱遥感　切入点：填图算法　出处：《成都理工大学》2015年硕士论文

【摘要】：遥感技术的发展和广泛应用已经使得人们不需要直接接触地表目标和区域就能获取相关数据,然后分析处理后得到所需信息。20世纪80年代以来,高光谱遥感技术的出现和发展让人们通过遥感技术观测和认识事物的能力又进步了一个台阶。高光谱遥感技术又称为成像光谱遥感技术,它是以光谱学为基础,,在400~2500nm波长范围内,获取光谱分辨率小于10nm的影像数据,其数据往往包含了空间、辐射和光谱三种信息。高光谱图像同时具有光学特性和光谱识别能力,其蕴含的丰富波谱数据使得地物的定量测量成为了可能,从而逐渐成为遥感领域研究与应用的前沿热点,并已经广泛应用在环境调查与监测、矿产探测、海洋大气土地动态监测、军事应用等广泛领域。高光谱遥感蚀变矿物填图方法是高光谱数据处理的关键,直接影响填图结果的使用精度和要求,文章主要围绕填图算法机理和应用进行了研究:(1)文章首先从地物光谱特征的机理入手,归纳了常见的蚀变矿物光谱特征形成原因,可根据地物不同的组成成分判断该种地物的光谱特征,并总结了常见的碳酸盐化、粘土化和二价铁等蚀变矿物波谱曲线。(2)详细介绍了当前常用的几类高光谱填图算法的概念和原理,总结了光谱特征提取的方法。通过实验构建模拟数据评估了光谱角填图法(SAM),光谱信息散度法(SIDM)和混合调制匹配滤波法(MTMF)的性能和精度。经对比分析,数据的噪声会对几种方法的填图精度产生较大的影响。因此,针对噪声较多的影像在填图之前需要进行降噪处理。(3)阐述了高光谱数据处理流程与主要方法,以研究区SASI高光谱数据为例,基于油气烃类微渗漏理论,提出了基于“特征光谱掩膜+MTMF”填图方法。实现了对研究区的烃类、粘土类和碳酸盐类等蚀变异常信息的提取,并有效的剔除了干扰因素。结合已有地质资料和野外实测光谱验证,取得了良好的效果。
[Abstract]:The development and wide application of remote sensing technology has made it possible to obtain the relevant data without direct contact with surface targets and regions, and then analyze and process to obtain the required information since the 1980s. With the emergence and development of hyperspectral remote sensing technology, the ability of people to observe and understand things through remote sensing technology has been further improved. Hyperspectral remote sensing technology, also known as imaging spectral remote sensing technology, is also known as imaging spectral remote sensing technology. It is based on spectroscopy, in the 400~2500nm wavelength range, the spectral resolution is less than 10nm image data, its data often contain three kinds of information, space, radiation and spectrum. The hyperspectral image has both optical properties and spectral recognition ability. The rich spectral data made it possible to measure ground objects quantitatively, which has gradually become a hot spot in the field of remote sensing research and application, and has been widely used in environmental investigation and monitoring, mineral exploration, Hyperspectral remote sensing mapping of altered minerals is the key of hyperspectral data processing, which directly affects the accuracy and requirements of mapping results. This paper mainly focuses on the mechanism and application of mapping algorithm. Firstly, starting with the mechanism of spectral characteristics of ground objects, the causes of spectral characteristics of common altered minerals are summarized. The spectral characteristics of the ground objects can be judged according to their different components, and the common carbonation can be summarized. The concept and principle of several commonly used hyperspectral mapping algorithms are introduced in detail. The methods of spectral feature extraction are summarized. The performance and accuracy of spectral angle mapping method (SAM), spectral information divergence method (SIDM) and mixed modulation matched filtering method (MTMF) are evaluated by constructing simulated data. The noise of the data will have a great influence on the mapping accuracy of several methods. Therefore, the process and main methods of hyperspectral data processing are expounded for the noisy images which need to be de-noised before mapping. Taking the SASI hyperspectral data in the study area as an example, based on hydrocarbon microleakage theory, a mapping method based on "characteristic spectral mask MTMF" is proposed. The information of hydrocarbon, clay and carbonate alteration anomalies in the study area is extracted. The interference factors are eliminated effectively, and good results are obtained by combining the existing geological data and field measurement spectra.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P627

【参考文献】